Four specific biostatistical inference problems related to cancer studies will be investigated in this project. A. Nonparametric or seminonparametric methods will be developed to analyze nondecreasing repeated observations over time with possible missing values. B. Nonparametric methods will be developed to estimate the survival curve under a particular dependent censorship. C. Under a randomization model, the poststratified and subset analyses will be studied based on restricted randomization rules. The group sequential methods (especially for survival data) will also be studied under this setting. D. Nonparametric methods will be developed to analyze a particular kind of nonrandom missing data.